The Periodicity Of Hundred-Year Solar Filament Archives Based On Time-Frequency Analysis Algorithm

There are various kinds of eruptive activities on the surface of the sun.Solar filaments,called prominences when they appear above the solar limb,are one kind of them.They appear as black stripes because they absorb the light radiated from the sun,which was the reason why they were entitled filaments.Filaments always appear above polarity inversion lines(PILs)of the photospheric magnetic field.More than that,the erupt of flares and the coronal mass ejections(CMEs)always accompany by the largescale changes of filaments.Therefore,researches based on filaments are of great significance to filaments and other solar activities.In the recent study on solar filaments,the National Observatory of the Chinese Academy of Sciences has proposed a new comprehensive dataset of solar filaments of100-year interval,which covers the data from five observatories in the world.The dataset extended from 1912 to 2018,which is relatively scarce in astronomical research.We use concentration of frequency and time to extract the time-frequency component of the data to analyze the periodic characteristics of the data.Compared with the synchronous compression wavelet transform,the concentration of frequency and time has stronger noise robustness,which is put forward on the basis of the synchronous compression wavelet transform.In this paper,we will introduce the work on this batch of data,mainly including the following three aspects:(1)Data preparation.We deduced the formula by which can convert filaments' pixel coordinates in the images to actual longitude and latitude.Then we realized algorithms with computer to transform the huge amounts of data and stored them in JSON files.In order to facilitate storage,we analyzed the data structure,obtained the main entities and the relationship between entities.We designed a relational database with reasonable data types based on My SQL,parsed JSON files,and imported files into the database.Finally,in order to excavate the scientific value of these data,we used Java language to write a network platform based on Browser / server mode to publish the database to maximize its research value.Through the system,users can realize multi-property query on the basis of demands,download data under multiple conditions,perform visual query information for one single filament.(2)Authenticity verification.By analyzing data in the database,the authenticity of this batch of data was preliminarily determined.By comparing the curves of filaments' monthly numbers and the international sunspot area,we found that the curve of filaments was roughly the same as that of the international sunspot area.Further research on the location distribution,yearly evolution with time and latitude distribution of filaments shown that dataset was basically reliable.However,there are too many error detections in the dataset,resulting in too many small-scale filaments.Before the periodic analysis,we plan to eliminate filaments whose area are below 150 mm.(3)Periodic analysis.We use application of synchrosqueezing wavelet transform(SWT)and well-focusing time-frequency analysis algorithm(Conce FT)to obtain the periodic characteristics of the dataset.By comparing the results,we find that: 1)the periodic components obtained by SWT are similar to those obtained by Conce FT,and there are also some deviations;2)Conce FT outperforms SWT on the noise robustness.The results of Conce FT are more reliable.3)The obtained periodic characteristics are:22 years magnetic period and 11 years Schwabe period with relationship of multiple frequency,the third harmonic component of Hale period,the first harmonic component of 11 years Schwabe period and some mid-term periods within the QBO.The results are consistent with the previous.